Table 2. Performance of ANN model and LR model in differentiating different histological grades.
| Models | AUC (95% CI) | Sensitivity | Specificity | |
|---|---|---|---|---|
| Training set | ANN-AP | 0.945 (0.865–1.000) | 0.980 | 0.962 |
| LR-AP | 0.805 (0.710–0.901) | 0.627 | 0.962 | |
| ANN-HBP | 0.975 (0.925–1.000) | 1.000 | 0.970 | |
| LR-HBP | 0.820 (0.724–0.917) | 0.783 | 0.788 | |
| ANN-AP + HBP | 0.953 (0.907–0.998) | 1.000 | 0.792 | |
| LR-AP + HBP | 0.921 (0.856–0.986) | 0.773 | 0.979 | |
| Test set | ANN-AP | 0.889 (0.804–0.974) | 0.913 | 0.882 |
| LR-AP | 0.777 (0.681–0.864) | 0.652 | 0.882 | |
| ANN-HBP | 0.941 (0.886–0.995) | 0.979 | 0.926 | |
| LR-HBP | 0.819 (0.742–0.895) | 0.883 | 0.722 | |
| ANN-AP + HBP | 0.944 (0.887–0.996) | 0.840 | 0.998 | |
| LR-AP + HBP | 0.792 (0.681–0.904) | 0.680 | 0.854 |
ANN, artificial neural network; LR, logistic regression; AP, arterial phase; HBP, hepatobiliary phase; AUC, area under the receiver operating characteristic curve; CI, confidence interval.